Sachin Dixit's Blog

There is one social media post that keeps doing round. It says that some NASA scientist claimed Sanskrit is most suitable language for AI.People laught at it as fake news. Here is link to 1985 article that started it all.
But why is this fact not underlined loudly you may ask. Multiple reasons.

First the fact that sanskrit has oldest codifed grammer ,some 3,4 BC are attributed to Panini who worte it. The main feature of this language is that it denotes same meaning even if you put words in any sequence. This happens via reflectivity .In simple term the pre-post fixes of noun applies to verb, adjective and derived forms. This is very neat preservation of semantic meaning. For other aspects of grammer this language has very logical,maths like rule.Which feels like a programming language.

Now, this aspect of sanskrit is given mention in intro sections of big books on compilers and pasrsers.Including the lagend MIT professor Noam Chomsky whose language theory formed basis of Natural Language Processing. But world has moved of from prolog and lisp to more sophisticated models of NLP. So the early theory of the field wont get mention in new papers ,there is no motive here. Same has happened to works on compilers.

Cut to today our early NLP libraries built on work of many researchers after Chomsky and relied on typical parse-toekn-interpret sequence.
The current champion of NLP ie BERT was celebrated for bi directional application of relationship which allowed it for better correlation of words in terms of their meaning. This feat actually derives from Sanskrit but its not an exclusive feature of sanskrit.So we dont have a reason to find prejudice.Nor can one person read all research papers where it might be mentioned.
Now the next phase of evolution of NLP is to move from sentences and arrive and comprehension at higher aggregate level.Like how you and me can understand poetry in all of its abstract metaphorical erratic flow.This calls for not only processing of language in terms of words-relation-meaning but also calls for iterative dive into parallel knowledge models and alternate meaning.Think of reading a satirical poem….
When the next production strength NLP winner, will be published it will derive from the reflectivity Sanskrit has; but its mention in the research paper will depend on the prior work the researcher has referred.
So,as much good the ancient languages is the way research methodology and citation works,we dont have to always suspect motives .

ps: I have tried to give very simplified view of NLP and Sanskrit here so experts in the field should pardon the simplified version of things.

Link to original paper of the images shown here : https://ojs.aaai.org/index.php/aimagazine/article/view/466

link to bert paper https://arxiv.org/abs/1810.04805

here is excellent article on timeline of parsing as technique : https://jeffreykegler.github.io/personal/timeline_v3

At the face of it ,its a fun language akin to brainfuck . That is main reason it is going viral now . But bhailang also highlights thee need to teach programming and computer science in native languages natively .
Each spoken language has different conceptual coverage and interrelationship .So when we teach a programming language (or any other science ) our main task is to translate these programming abstractions to their best conceptual match in the mental language of the student (every one has a thinking language).
One simple example is Indian usage of words like “relate to” or “adjust” .These are traditionally alien concepts to us .But we have kind of figured out what they mean in essence and we use them with fellow Indians. A native English speaker will find it amusing tough .But indians get it .That’s how spoken language work. They represents a unique world of psycho-conceptual model that works on known conventions .
Coming to programming . What do we mean a=1 . Does a contain 1 or does it represent or does it refer or is it equal or is it assigned . Your answer will depend on the programming language .Your life is even more tough if you know multiple programming languages .

Who then makes sure that such thing is aptly conveyed to student in his thinking language .Consider another case , who do we really mean when we say a class abstracts another ? Can you explain that to me in your native language succinctly ? There are many such concepts in software which we haven’t consciously translate + conveyed (trans-convey) in our thinking languages .But in our mind this happens anyway ,unconsciously whiteout any formal check on correctness of it .

Few more examples: is specialization of , function is applied to ,is executed ,represents edge case , manifests as, multiparadigm or imperative language , normalization ,process communicates, async processing , stacked vs queues ,modular ,layered ,optimized …..

Of course all of us vernacular folks and their teachers have “kind of figured out” what all this means .But “kind of figured out” is not really the scientific way of learning nor can it guarantee accurate outcomes .
Bhialang with its popularity , raises this very question of trans-conveying of software concepts !

Check it : https://bhailang.js.org/

the great resignation wave has resulted in great salary inflation in India. For long multiple of 1.5 to 3 was where the mean lied . Now its very common for younger professional to claim 3 to 5 multiple. The skew is higher at less experience but that’s common thing in stats 🙂

This is not first time

But this not first such rerating ,we had similar events in last 2 decades .One in 2006 era and other one in 2012.In both cases the salaries were rerated .But in the years to follow we had very slow growth with passive push back from companies . This brings us to the discussion of brutal appraisals we will have in 2022-2023 era.

The hard math behind salaries

Every company whether they have a flat agile team or pyramid teams have revenue value assigned to their people. For services its driven by billing and is more obvious. For others its unspoken but such numbers do exist. Once this rush is over its natural that product owners and likes will do the hard math[for most companies IT is cost center ;for some its key differentiator ]. That’s when the hike, retention and appraisal discussion of next year are expected to be brutal.
The case might be easy if you’re working at internet giants or flush with cash unicorns. But for a vast majority these multiple will pose a moment of truth.

How to ride the tide

My advice to younger folks is to increase their talent multiple as much and as fast as possible. There are few skills in software like design, architecture, security, networking that are hard skills and they are function of time and exposure. But for professionals working in pure play language programming or framework programming its easy to level up in short time with practice. It is also important to note that pure play language or framework programming is also where people replacement is easy (its very easy to hire freshers in bulk and wait for them to mature ,say about a year). In  case of framework programmers the sheer speed of changes in their ecosystem means that such salary gains come with a time to live value attached to it.
So this great salary rerating is also call to action for making the individual difficult to replace :). Difficult in the sense of depth ,breadth, command over craft and also future proofing. Or else the same economic principle or supply demand that apply to goods can easily come hitting hard.

so level up, invest in your self fast.

( pasting below a skill map for anyone to chart their training journey in Indian software industry ,its something that I have been following for years and my resume has been styled as such 🙂 .I don’t claim this to be path breaking stuff but over years my Jrs have benefitted from this so there is no harm in having a look 0

Here is link to the xls i created ,if the image isn’t clear

https://docs.google.com/spreadsheets/d/1l5TBR76nEh-dUXc4FL1e95b2C9hadzn2vGiTb4MrSrg

Every morning when I wake up I get into panic till I start my laptop and login . This is shaped by workplace in distant past and old bosses who was proud of being very particular in tracking 9.15 hours . And then i become senior member of my team and it didn’t matter .But the panic still remains.

In other variation I feel that my boss will fire me today. There is nothing in my work history to support this, but the fear is real.

Back to work plans

Now that we are talking of return to office in 2022 it is important to note that emotional well being might be the key trend of the decade . The 2 plus years of pressure cooker experience we all had-still having has impacted us in many ways . and we don’t even know where and how of it. To begin with we need to accept that fears are not rational. We need to also move from cinema inspired list of mental issues to common man variety of emotional issues ,like i gave above . Which also means that the working mental health makes it very derogatory and un approachable .A more humane emotional wellbeing would be better view point to see it all .

What awaits us at workplace now

In terms of kind of emotional issues we might see , one need to move away from laundry list of ADHA,OCD, divorces and think about grief, dread , fatigue , distortion . These are more likely to show up at workplace as they work silently in the background .The last thing to note is that covid19 was like experiencing a car accident for everyday in last 2,3 years . The impact will be generational .It will take decades of work to get back to cheerful ,focused, motivated individuals that can operate in teams well . This also means the word productive is bad word, we need to now embrace healthy work as the paradigm .

The shift we need to make

One major point is about training and support .The default tendency of organization is to offer meditation or yoga session and email an counsellor helpline. However post covid world will need all of us at the floor to have more empathy , purpose and human connect .It is like moving back from industrial to tribal era of where your group was your main pillar of strength .I wonder how many companies can claim this in a healthy way . However, if you’re looking for arthritis relief, you might try CBD for arthritis relief your health concern.

The last thing is to also acknowledge that everyone in your hierarchy and business chain has been affected emotionally and are trying to balance themselves .It would be wise to think that we are now a gathering of teenagers helping each other to make sense of the world and unknowingly also decide how the generation will shape up. This not to sound alarm but emotional issues can run deep and take lifetime . I hope we and our workplaces are ready for it .

To kickstart you thinking , imagine that you join back at work and find one of you mate heads on desk style sleeping . How do you plan to react ? thats our starting line now …

The log4j incident spoiled the weekend for many people, including myself 🙂 . There was similar incident years back when some basic library around padding(read more :left-pad) underscores in nodejs ecosystem was taken off by its maintainer . With open source -contributed modules proliferating across our software stack organizations now need to be mindful of their software supply chain .This is in line with how covid brought the same realization in manufacturing .

Hard fact about digital supply chain

If you are building a spring-angular kind of application , the long list of nested dependencies can be overwhelming .Now multiply that for the server stack you are running on . Below that the OS or cloud vendor would have some such . And below that your Hardware provider would have some . Orthogonal to are your logging solutions-gateways and downstream services . This is like movie inception .A world within a world -a world along side another .Now that we are aware of the rabbit hole ,lets go inside .

Need for mindful package dependencies

When we use library like spring or some apache solution like spark , the main focus is either on license model they have or feature comparison with an alternative .These are building block of our solutions now .Time has come for companies to be actively contributing to them .This could be of course in donations of money or developer time. For companies who dont want to have direct touch point ;new companies can come up to serve this need(check github sponsors ). Such a model already exist when it comes to corporate social responsibilities CSR mandates

Active governance of the stack

More important than the contribution is governance . We know multiple consortium like Foundations that exist via Apache , Eclipse , CloudNative or even old school JCP . Typically its software product vendors that show up here and most of the noise is around features . Sometimes politics 🙂 . There is a need of governance body or bodies that focuses solely on well being digital supply chains .It would be lightweight body focusing on compatibility, security and selection -retirement of dependencies . The mavericks might not like the idea of someone telling them which library to depend on . But such a choice is made by all developers where they consider health of a repo -popularity etc . Moreover remuneration of the maintainer of dependency module is one important aspect . When a disruption similar to log4j had happened with openssl , people were shocked to know about the sole maintainer of this library and how much he was paid ! ( read details here) What if such a sole contributor is sick when zero day issue are found or decides to retire . We are not even thinking of digital equivalent of repo inheritance ! About time we do

Possibility of espionage and cyber warfare

This might look alarming but this is needed to complete the converge of topic . When your code depends on a complex web of dependencies , it is a soft area for enemies to hurt . I am not aware of studies done for this topic (eg china study) . But it is not unthinkable for bad actors to get into role of contributor or some role into these repo . Once the actors are in they can possibly inject deep level backdoor or even a time or event based memory crash is good enough . I don’t want to paint a dark picture here but we have left the fences open for long here …

Politics can get in too

The nodejs issue cited above also has developer-corporate power tussle angle to it . There can always be some conflict between maintainer of a contribution module and the host body . Leaving aside the many ways this can happen , what fall back mechanism we have in such case ? Say a contribution module suddenly is taken off or expelled but your build pipelines have already married to this module , now ? We need a governance here .

At the same time one need to be aware of developer activism . Given that we have across globe developer communities it is not impossible that developer stage some sort of satyagraha ( refer Gandhi) and take down their repos . Or there is some huge anti amazon or anti someone sentiment and they change their licenses .

A more apparent one a country like china or someone against your country taking some action throwing your supply chain in disarray .

Stack Overflow here

It is little bit of expansion but given that a lot of code is copied from stack overflow these days , it is one hidden supply chain we have in our stack . Who owns a deep bug in such copied code ? or a recommendation around using some library . There can be many internal process checks here but stack overflow or its equivalent content site is part of the radar now .

The rise and rise of salary offers

If you read popular news articles in India about IT, your would get a picture that everyone is changing jobs .Lately some media house have also started talking about the hikes and how big they are .A lot of these construe that due to covid19 people dint get hikes and now the jump ,or something on these lines .

Enter LinkedIn posts and you see lot of youngsters posting about how they got some offer and they manger to outgrow that with counteroffers .The percentages are huge .

At the same time ,the business newspapers and quarterly results hardly talk of huge contracts coming their way. How does then one connect this and form the coherent picture ?

The dollar peg of Indian IT salaries is off now

In a globalized business it is hardly a surprise that exchange rates come into picture when deciding employee salaries .However in the initial decades it was the IT services that defined job market. It also brought billing rates with it .The rates were ,generally linked to roles and designations . It also meant the hikes and promotion were aligned to annual results and big hikes came your way when roles changed . This is so fundamental to us that our hiring-retention discussion still center around this .

It is not an exact year but around 2008ish is when the captive unit movement started in India. This was first instance of that billing rate foundation was shaken .More or less .But the exchange rates still remained in background .

By next 10 years .To put a precise date , around the time Flipkart and Ola happened in India we had mass scale recruiters that the foundation of dollar peg was taken off. The more and more build for yourself software companies open here, the more prominent this trend is .

Covid19 or Buzzing startup : Where is the tipping point ?

Of course the startup scene in India is maturing .Do we then conclude that this alone is causing the scale of salaries in job offers to rise suddenly ? A buzzing economy does cause good hikes in salary , we saw that in 2003-2007 . So does post social shocks , like the hikes in 2011-2014 era. But that does not explain the suddenness .Nor do LinkedIn posts mentioned above say that everyone is jumping to startups .

It would be useful to consider other social-human aspects that covid19 brought to us.

1. Onsite is less lucrative : The main hidden attraction in services companies was onsite posting .Covid 19 has made it look less attractive or even bad choice . So any market worthy candidate is now forced to discover his native-local worth .
2. Subtle Social pressure is off : In absence of a survey , it would be difficult to guess the contribution of office atmosphere into long stays at companies .But this is important factor why people stay .Your friend circle ,the lifestyle keep people hooked .In some cases an inspiring or imposing boss contributed . In some cases your yesterdays bug or mistake made you push thinking about career to next day. And the main factor is formal arguments .There are many formalized workplace related arguments and anecdotes about satisfaction ,purpose and all that worked in formal work place .A lot of them meant different thing when you sit at home in lockdown. In some cases people told me they laughed at these stuff now.
3. Family time reality check : There is big value that we attribute to continuity. This is one invisible force that keep people into same circles .In many cases semi urban rural families thought greatly of the work, workplace, companies .Working from home brought the reality check .The long calls ,shouting bosses and hours of hunch back coding on sofa looked different to them .Of course some must also be touched by care shown by companies during covid times .But the rosy picture that you have a IT-working kid that is very attractive in marriage market is realistic . The cost of the salary is obvious to them now. In some cases either families or individuals also got space and time to consider alternative careers .

All of above factor lead to an individuals who are well reflected about basics , bolder about their demands , gusty about the bargains demanding sharper differentiation . The LinkedIn post you see about counter offers are from such people who have market value !

The aftereffect we need to be aware of

My old friend the other say forecasted that this trend will not sustain .As of there is some alignment in poverty and forces of nature .One even said that later these companies will fire them or give 1 % hike .Any such comment ,reasonable or not misses the fact that hiring companies are also rational actors.

When we postulate that dollar peg of Indian IT salaries is off ,we must also consider that the talent peg is now global . Increased remote work and more startup had to have this effect. So the net result is that not only salaries but the skill demands are also catching up with global peers (emphasis is on skill demands as opposed to skill levels).

This is not to say that local talent was lesser .It is just that both factors are in sync .Looks like they will remain in sync. This needs some consideration by career minded people .

When we talked of billing rate dominated -pyramid workplace .We dint talk of a work place where skill levels where matched to client needs .In many cases the skill range was narrow ,in some cases it was not latest and in some cases it even allowed people to hide behind the pyramid without having good skills .

The movement from waterfall to agile in many cases exposed this .It is thus no wonder that agile and reskilling became buzzwords around the same time in India .How ever this far the the skill demands and salaries where not rising in sync.

I would like to postulate that now they are .This means that not only the skill levels will have to be globally competitive but also that the onus will now be more on individuals to stay on the edge of competitiveness .Proven and impressive hands on skills are already the norm in Indian IT hiring .This might also get into zone of industry leading talent levels .If the salary curves are to sustain and grow across years , aspiring candidates might be also judged on their tech-though leadership . Say open source contribution or patents or research papers would become norm. Ones teaching technology on professional platform or with books on their names .Even difference of opinion on established technology norms count as positive !

All of this is actually is good news for the community as well as the industry and the individual .We are looking at fitter techies who are well paid, can lead the agile trains, deliver industry leading work. Its win-win-win .

The only looser might be candidate who think that they are outsmarting the industry by squeezing the “hot market”.

In reality there is no hot market .All we have is market demanding top skills and able and willing to pay top rupee for it .Have good next decade Geeks !

Of course, each business newspaper is telling you that attrition levels are high in so many IT companies in India .One gets to see huge percentage numbers cited in such news . But very few of such news present the the details of such movement leave alone the causes . A few article and posts did mention that last year due to covid the hikes were less or nil . But does that qualify such big jumps in attrition .

A deeper discussion with the folks exiting tells diff story . In some cases the women have realized ,after staying with kids for a year that their IT job is not that “needed”. In some cases the men have questions need of an IT job itself for livelihood .In many cases a lot of folks I talked to had farm or fathers business as a fall back .And the covid wfh allowed them to consider going back seriously .In some very sad cases ,friends had causality or very bad episode related to covid which made them question the need of city job away from near ones . A few youngsters told that a year at hometown they would do a start up because a formal IT office doesn’t excite them now.

Of Course mine is not a detailed study with lots of data .But it should tell that there are some fundamental shift in Urban-9.15hours-IT jobs that were seen as the thing in India for past decades .One might read some US based publication and call this quit job movement but that would be superficial .I would rather call this as reverse migration or some sort of coming full circle on Urban migration as option and IT job as the best lifestyle choice .Both have been challenged by covid. So next time someone says the people are jumping jobs for money , it might be t he symptom but the causes might be disillusionment .Oh and by the way you where thinking of allowing your team to work 2 days a week from home, right ? Welcome the HR challenge of the decade .

The paradigm has ,in this case ,actually shifted and we don’t exactly know where 🙂

Lead question : Why API First

Ask a product manager what is the different between made or order software and a product .He will tell you how product is general purpose for the given domain (not generic tough ).How it captures a broad set of use cases from the domain .How it is flexible in adopting to diff combinations of rules that are included in these use case .How it is future proof and is ready for future changes in domain .How it benefits from vast experience from we have in the domain/customers /consultancy . It goes on .

Yet when it comes to day in life a of developer he is often made to code first , code just what he has been asked for and ship it . It is obvious that a start up will only put coding effort into what sells but should that rule apply  to the very definition of the product as it is conceived  ?Is defining a product/service/roadmap same as coding it ? Such gap between what is “Seen” by product and what developers are made to do are solved by evolution read software design . API first is a prime example of being comprehensive and thus evolution ready yet riding the demand wave frugally .

Business scenario for  API first

• Integration /Consumption : Software need to be integrated by different parties as the success happen .Would you like to struggle to meet demands of this success come your way ? or would you like to be ready for it with some top-up effort .The answer is in thinking API first . Once you  think of well  thought of APIs  gluing them become easy and future successes becomes faster . This could mean integrating in or integrating out ie consuming (and monetization ).
• Aggregation/federation  : The internet economy of aggregation made famous by amazon/uber has made clear a case for softwares co-operating with each other .This increases market reach and revenue .However aggregation economy is volatile and competitive . One need to be ready at shorter notice than typical integration projects .The integrations also need to be robust for the parent party (and their customers ) to trust you .Another flavor could be federation of services amongst equals . A well thought of API on both sides make such integration easy-fast-reliable .
• Extension  : Acquisition ,Saas enablement  ,inhouse aggregation(called integration ) or even customization are typical cases of base capability of software being enhanced by extension .We may thinking of moving from web to mobile to chatbots/alexa . We may also consider taking base software and enabling that as multitenant offering .  While these extensions are value positive for every one they need be very liberal in capability but restrictive in scope .This help us keep sanctity of base software  yet allow creative but reasonable value add . An API layer will clear laid out  extension helps .

All of these driver point towards one fact that business that are built with well defined contracts stand to benefit from this opportunities as opposed to software that are inwards focused and need extra work for any opening up .While this discourse underlines that API first help us with readiness in terms of external forces , the same game plays out within the teams also .It is just that inter team play is seen in the light of following themes

Advantages of API first approach

1. Domain fitment : The tendency for Business analyst and developer to grab a piece of functionality under discussion and “just code it” .This has huge chance of teams creating code that is few layers higher in domain and miss out in fundamental building blocks .An API first approach also forces teams to think bottom up inside the domain where they are made to think of the “built up” of the functionality .This ensures that the domain of software in considered in totality before team chooses to pick up one slice of it and code it .
2. Parallelization : API also stand for clear contracts .So it allows teams to do more parallel work as api-first minimizes risk of large scale integration of parallel work streams .Productivity and time or marker related gains follow .Agile ie .
3. Testing : Since APIs are laid our clearly the function or security test team get more and enough time for deep thought testing approaches at all levels of test coverage .
4. Functional resilience : Since API commits to a final contract it gives a robust functional capability which is strength .At the same time there are many internal implementation details are that are open to amendments which adds to overall resilience .Think of a case where old rule based API is replaced with new Machine learning based API ,while the API is same the quality of algorithm is smarter .
5. Evolution : A well laid out contract also facilitates future evolution easy .It could be new version , a customized or localized flavor which offers changes in the same neighborhood as original older version of API (hence evolution ) . This however needs a better governance put in place .
6. Tech Debt repayment : A well defined contract also allows decoupling between team producing it and team consuming it . This allows the team to keep on changing internal technical detail of their API healthy and repay tech debt without disruption .
7. Other than code needs : There are many other than code needs like performance ,build , documentations , availability  , monitoring ,audit ,security which are get into motion after the software is committed to repo .API first lays out clear shape of the software to be and allows these team to think thought early on and facilitates these other than code needs at much deeper level of engagement .

Pitfalls and background of API First

API first is often confused with webservices .This often leads teams into thinking that if they are not exposing any url style webservices to anyone they are free .A better term for API-first could be contract first (but that’s very technical world so API is used ). This means that no matter whether one is creating piece of software that is consumed by outside world or other teams/developers in your company ,so long there is interdependency one need to commit to clear contract .And in order to see the gains of agility etc one need to commit to this contract first .This also means that all layers of software from external facing UI -webservices to components-frameworks in middle layer till data-integration layer one need to think  of defining clear contract .To that effect every component is an API that is producing or consuming other API . Even so-called cross cutting layers like logging are mater of laying out clear cut APIs.

The concept of defining clear interfaces in now new to programmers . We often talk of specifying clear interfaces and programming to interfaces .However this good practice start at individual program and ends at design patterns . May be it is seen as something only serve side programmer do . But , as we approaches more layered , more distributed(teams as well as software) , more integrated and more reusable (think open source ) , more componentized and so on the value of clearly laid out contracts between these seams is apparent . API first is this a mainstream realization of this fact , tough people confuse API with webservices only .

more reading ,Bezos Amazon API Manifesto and link to Yegge of Googles Rant : https://gigaom.com/2011/10/12/419-the-biggest-thing-amazon-got-right-the-platform/